Objective: Patient identification (ID) bands are an essential component in patient ID. Quality improvement methodology has been applied as a model to reduce ID band errors although previous studies have not addressed standardization of ID bands. Our specific aim was to decrease ID band errors by 50% in a 12-month period.
Methods: The Six Sigma DMAIC (define, measure, analyze, improve, and control) quality improvement model was the framework for this study. ID bands at a tertiary care pediatric hospital were audited from January 2011 to January 2012 with continued audits to June 2012 to confirm the new process was in control. After analysis, the major improvement strategy implemented was standardization of styles of ID bands and labels. Additional interventions included educational initiatives regarding the new ID band processes and disseminating institutional and nursing unit data.
Results: A total of 4556 ID bands were audited with a preimprovement ID band error average rate of 9.2%. Significant variation in the ID band process was observed, including styles of ID bands. Interventions were focused on standardization of the ID band and labels. The ID band error rate improved to 5.2% in 9 months (95% confidence interval: 2.5–5.5; P < .001) and was maintained for 8 months.
Conclusions: Standardization of ID bands and labels in conjunction with other interventions resulted in a statistical decrease in ID band error rates. This decrease in ID band error rates was maintained over the subsequent 8 months.
In 2007, the World Health Organization recommended a standardized approach to patient identification (ID) with a patient ID band that contained at least 2 patient identifiers to decrease the risks of medical errors from misidentification.1 Limited data exist regarding the number of patients harmed by incorrect patient ID, but multiple studies have documented harm related with laboratory testing errors, incorrect blood transfusions, and medication errors.2–7 Barriers to use of an ID band include behavior and culture, process variation, and inaccurate patient identifiers.1
There are published adult and pediatric data on error rates of ID bands and quality improvement (QI) strategies to decrease these errors. The largest of these was the College of American Pathologists Q-Tracks study at 217 adult institutions; it demonstrated an initial ID band error rate of 7.4%, which was decreased to 3.1% with monitoring and best practice sharing.8 Hain et al9 published the first pediatric-focused study on ID bands at a single site; they found an initial ID band error rate of 20%, with sustained improvement to 2.6% with awareness, staff surveys, and education. Building on the experience of Hain et al, a subsequent 6-site pediatric learning collaborative showed an initial average ID band error rate of 17% across institutions.10 The collaborative was able to achieve a 77% relative reduction in ID band error rates to 4.1% by using a combination of educational initiatives, staff surveys, and changing ID band type for neonates.
In 2010, accurate patient ID was selected as an organizational priority at our institution. Children’s of Alabama (COA) is an urban pediatric tertiary care hospital. At the time of the study, it was a 290-bed facility with 19 pediatric ICU, 26 step-down ICU, and 42 NICU beds. Based on our voluntary reporting system, there were 1487 patient ID reports between January 2010 and December 2011, which included ID band and surgical, laboratory, and medication errors of all safety severity levels. Patient ID errors were found to be the root cause of 19 serious safety events in the same time frame. There were limited data, however, related to institutional patient ID processes and ID band error rates. With the knowledge that other pediatric institutions had initially encountered double-digit ID band error rates, a multidisciplinary team was gathered with a global aim of improving patient ID and a specific aim of decreasing ID band error rates by 50% within a 12-month time frame. The team included the chief medical officer and chief nursing officer as champions and the medical patient safety officer and risk manager as sponsors. Front-line team members included hospitalists, nurses, unit clerks, the ID band vendor, and staff from the patient registration and the purchasing departments.
The University of Alabama at Birmingham institutional review board approved this study. Because patient ID is a crucial aspect of daily patient care and no patient identifiers were collected, we did not obtain written informed consent from patients or families. The data collection team included 5 staff nurses, an advanced practice nurse, a medical student, and 3 hospitalists. The Six Sigma DMAIC (define, measure, analyze, improve, and control) QI model was used, and the ID band processes and errors were defined at the beginning of the project (Appendix Figs 1 and 2).11
Members of the data collection team conducted weekly unannounced observations of the 13 inpatient nursing units from January to April 2011 for a total of 12 consecutive weeks. Observations were performed at different times of day and week, including weekends and holidays. The 13 inpatient nursing units included all medical/surgical acute care beds and the PICU and NICU. The inpatient psychiatry, burn, and stem cell units were not audited because of different patient ID and isolation policies related to the nature of their patient population. Standardized feedback to the nursing staff (Appendix Fig 3) regarding the ID band errors on the unit immediately after an observation was performed from the onset of data collection.
Each member of the data collection team was randomly assigned 2 units per week and used a standardized tally sheet (Appendix Fig 4) to record the data. Data were only collected for those patients who were physically present in their room at the time of the audit. The data collector audited patients for presence of ID band errors, which were defined as follows: (1) an absent ID band; (2) an incorrectly placed ID band, which was present but not physically on the patient (ie, taped to the bed) or illegible; (3) a noninstitutional or noninpatient ID band; or (4) an ID band with inaccurate name, gender, or date of birth.
The name, gender, and date of birth were verified with the patient (if≥10 years old and competent) or a parent/caretaker. If neither were capable or available, we verified the patient information with the nursing staff based on information from the chart. From observations 7 through 14, data on ID band styles were obtained when there was an ID band error but then expanded to every patient on observation 15. Targeted data on ID band styles were collected for 2 weeks (between observations 14 and 15), and these data were not included in the final analysis because data were only targeted to ID band errors related to absent, incorrect type, or placement of ID bands.
After the data collection period from January 2011 to April 2011 (observations 1–12), observations were not performed weekly but periodically over the remainder of the study period. Data collection was otherwise performed in a similar fashion, with all 13 inpatient nursing units audited in a 1-week period. Observations remained unannounced at different times of day and week.
Data were entered into a spreadsheet by a member of the team and were reviewed by another member of the team for accuracy and validity. The data were analyzed by using Minitab 16 (Minitab Statistical Software 2010, Minitab Inc, State College, PA). A χ2 analysis for ID band error rates was performed for preimprovement (observations 1–16) data compared with postimprovement data (observations 17–26). A χ2 analysis for differences in nursing unit acuity and preimprovement and postimprovement ID band errors according to ages was also performed.
From the onset of data collection, our first intervention was feedback to the nursing staff after each audit of ID band errors on the unit. This intervention was performed so nursing staff could correct the ID band error and to increase awareness of the institutional policy and importance of ID bands. A second intervention during the initial data collection period was an educational conference on the topic of patient ID and ID bands. Based on information gathered from the first 14 observations, the first targeted intervention was the dissemination of institutional ID band error rates to nursing, physician, and administrative leaders in QI meetings. This led to support by hospital administration to standardize ID band styles and labels. Based on data from targeted data collection, we selected 1 neonatal-style and 1 child-style ID band. We then collaborated with our ID band vendor to select appropriate sticker labels containing the printed patient information for the child and neonatal ID band. These printed labels were then marked with an arrow designation on the label sticker sheet to clarify to staff which label was appropriate for each ID band.
After standardizing the ID band styles and labels, the next intervention was making the institutional and nursing unit data more transparent to the staff while providing education on the new standardized processes for ID band styles and labels. We educated staff on the policy for placement of ID bands at the time of admission for patients who were admitted from our institution’s outpatient areas. Nursing unit–specific fliers (Appendix Fig 5) were created and distributed to each nursing unit monthly with each unit’s ID band error rate and the institutional ID band error rate in graphical and written formats. The fliers also served as an educational tool on the new ID band styles and labels and instructed the staff on actions to take if a patient had an ID band error.
There were a total of 4556 patients audited from 26 observations over an 18-month period. Each observation consisted of 13 nursing units with the exception of 1 observation (observation 7) in which 2 units were erroneously audited twice and 2 units were not audited. Each observation had between 136 and 206 patients. The data table for observations is shown in Table 1.
The preimprovement data collection period was composed of the first 12 consecutive weeks of observations and 4 subsequent observations. The average ID band error rate for these 16 audits was 9.2%. The postimprovement average ID band error rate was determined once there were 8 points below the calculated mean and began at observation 17, which was 9 months after the project was begun.12 The postimprovement average of 5.2% was significantly lower (95% confidence interval, 2.5–5.5; P < .001) This finding is illustrated in the statistical process control chart in Fig 1.
Data on ID band styles and ID band errors were collected over a 2-week period between observations 14 and 15 (Table 2). Table 2 also describes the physical characteristics of the different ID band styles, and these styles are shown in Fig 2. The ID band styles without clear coverings (2, 4, and 6) were eliminated as options because the cover prevented printed information from washing off. Style 5 had a narrow design that resulted in difficulty fitting a sticker label. Two styles of ID bands, style 1 for neonatal and style 3 for child, were selected because they were the most commonly used and both had a lower ID band error rate than the preimprovement average.
The most frequent ID band error overall was that the ID band was not on the patient. We differentiated when an ID band was not on the patient by collecting data regarding an incorrectly placed ID band (ie, taped to the bed) versus an absent ID band. During the preimprovement period (observations 1–16), ID bands not on the patient (incorrectly placed or absent) accounted for 73.6% of errors (Fig 3). After improvement, ID bands that were not on the patient accounted for a similar percentage of errors at 80.7% (P = .23). Over the entire period of data collection, acute care patients and non-NICU patients had similar ID band error rates (6.0% and 6.2%, respectively); however, the NICU ID band error rates were lower than rates for acute care patients at 3.2% (P = .04). Although there was a decrease in ID band error rates in all age groups between preimprovement and postimprovement periods (Fig 4), only infants had a statistically significant decrease in ID band errors (P < .001).
We present our QI experience in which the decrease in ID band errors required standardization of the ID band styles and labels in conjunction with educational initiatives and institutional and nursing unit dissemination of ID band error rates. Although this was a single-institution study, thousands of patient audits were performed over an 18-month period. Our preimprovement patient ID band error rate average was 9.2%, which is lower than previous reported initial pediatric ID band error rates.9,10 There was a decrease in the ID band error rate to 5.2% in 9 months, with a sustained decrease over the subsequent 8 months. Our specific aim was to decrease ID band errors by 50% in 12 months, and we achieved a 43% decrease in ID band error rates when comparing the preimprovement and postimprovement averages.
The preimprovement period demonstrated wide control limits (0.0–18.5) shown in the statistical process control chart in Fig 1, which indicates a large amount of variability in the process.12 An illustrative example of this variation was the 6 different ID band styles shown in Fig 2 used at our institution at the beginning of our study. In the analyze phase of DMAIC performed throughout the preimprovement period (observations 1–16), the average ID band error rate remained unchanged despite ongoing nursing feedback regarding ID error rates. The decrease in ID band error rates with awareness and educational initiatives that was noted at other pediatric studies was not seen at our institution.9,10 This finding may have been related to an inadequate dissemination of the importance of ID bands, different institutional cultures, or the variability in our ID band processes.
Changing the ID band styles and labels was an intervention not anticipated in the define phase but was performed when the variability in process including the ID band styles and labels was noted. Based on targeted data collection (Table 2, Fig 2), a child and neonatal ID band style was selected based on usage, ID band error rate, and physical attributes of the ID band. The low usage and physical characteristics of the other styles were judged to be inferior to the selected ID band styles 1 and 3. After standardization of processes, the postimprovement period demonstrated a decrease in ID band error rates with narrower control limits (1.5–8.9), indicating a less variable process.12
As in previous studies,9,10 ID bands not on patients (incorrect placement or absent ID band) was the most common type of error in our study both preimprovement and postimprovement. As seen in the pediatric ID band quality collaborative,10 we observed differences in ID band error rates among nursing unit acuity. In contrast to this study, we found lower ID band error rates within our NICUs compared with acute care settings. We also noted differences in ID band errors among different age groups (Fig 4), although these were not statistically different in the postimprovement period. Infants are the most common inpatient age group in the hospital, and there was a statistically significant decrease in ID band errors among infants in the preimprovement and postimprovement periods.
There were several limitations to our study. This was a single-site study at an academic institution, and the results may therefore not be applicable to other institutions. We did not audit patients from the stem cell, burn, or psychiatry units. If the patient was not in the room or was undergoing a procedure at the time of an audit, it was at the discretion of the auditor to reattempt to collect data from the patient.
As more institutions use electronic patient ID technology such as barcoding and radiofrequency ID that incorporate the ID band, further research will be needed to assess what a consistently achievable ID band error rate is and the cost/benefit of continuous QI efforts to sustain lower ID band error rates in pediatrics. In the College of American Pathologists Q-Tracks study of 217 adult institutions, Howanitz et al8 found that the best performers (top 10%) were able to maintain wristband error rates at 0.2% to 0.3%. Only 1 institution, however, in the 6-hospital pediatric collaborative was able to achieve similar rates (0% error rate in the last month of the study), but they found that all sites except 1 made improvement.10 Although we did achieve sustainable decrease in ID band error rates, continued efforts at our institution to maintain and improve the ID band error rates are needed.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
- emergency department
- quality improvement
- Patient Safety Solutions
- Valenstein PN,
- Raab SS,
- Walsh MK
- Morrison AP,
- Tanasijevic MJ,
- Goonan EM,
- et al
- Serious Hazards of Transfusion
- National Health Services
- Hain PD,
- Joers B,
- Rush M,
- et al
- Phillips SC,
- Saysana M,
- Worley S,
- Hain PD
- Rath Strong
- Wheeler DJ
- Copyright © 2013 by the American Academy of Pediatrics